Abstract:

In current streaming application distribution system, the client loads application resources on demand through streaming loading, so the client needs to frequently access the remote server, which causes problems such as server overload, increased traffic consumption and slow application startup. Aiming at the above problems, the cache and integrated user behavior prediction strategy A-RBFS（Adaptive Recently Behavior Frequently Size） of the streaming application distribution system is proposed. The strategy considers the factors such as user usage behavior, client state and application size, and adjusts the total size of the cache space according to the size of the remaining storage space of the client. The implementation results show that under the same conditions, the cache replacement strategy is significantly better than the LRU and LFU cache replacement strategies.

Key words: streaming application, user behavior, cache strategy, load on demand

摘要：

在目前的流式应用分发系统中，客户端通过流式加载的形式按需加载应用资源，因此客户端需要频繁访问远程服务器，从而导致服务器过载、流量消耗增加、应用启动缓慢等问题。针对上述问题，设计了流式应用分发系统的缓存及综合用户行为预测策略A-RBFS（Adaptive Recently Behavior Frequently Size）。该策略同时考虑用户使用行为、客户端状态和应用大小等因素，并根据客户端剩余存储空间大小调节缓存空间总大小。实验结果表明，在同等条件下，该缓存替换策略明显优于LRU和LFU缓存替换策略。

关键词: 流式应用, 用户行为, 缓存策略, 按需加载